development/plasticity/repair … · 2006-11-21 · zation of nkcc1 (moore-hoon and turner, 2000),...

Development/Plasticity/Repair

Oligomerization of KCC2 Correlates with Development ofInhibitory Neurotransmission

Peter Blaesse,1 Isabelle Guillemin,1 Jens Schindler,1 Michaela Schweizer,2 Eric Delpire,3 Leonard Khiroug,4

Eckhard Friauf,1 and Hans Gerd Nothwang1

1Abteilung Tierphysiologie, Fachbereich Biologie, Technische Universitat Kaiserslautern, D-67653 Kaiserslautern, Germany, 2AG Elektronenmikroskopie,Zentrum fur Molekulare Neurobiologie, D-20251 Hamburg, Germany, 3Department of Anesthesiology, Vanderbilt University Medical Center, Nashville,Tennessee 37232, and 4Neuroscience Center, University of Helsinki, FIN-00014 Finland, Helsinki

The neuron-specific K �–Cl � cotransporter KCC2 extrudes Cl � and renders GABA and glycine action hyperpolarizing. Thus, it plays apivotal role in neuronal inhibition. Development-dependent KCC2 activation is regulated at the transcriptional level and by unknownposttranslational mechanisms. Here, we analyzed KCC2 activation at the protein level in the developing rat lateral superior olive (LSO), aprominent auditory brainstem structure. Electrophysiology demonstrated ineffective KCC2-mediated Cl � extrusion in LSO neurons atpostnatal day 3 (P3). Immunohistochemical analyses by confocal and electron microscopy revealed KCC2 signals at the plasma mem-brane in the somata and dendrites of both immature and mature neurons. Biochemical analysis demonstrated mature glycosylationpattern of KCC2 at both stages. Immunoblot analysis of the immature brainstem demonstrated mainly monomeric KCC2. In contrast,three KCC2 oligomers with molecular masses of �270, �400, and �500 kDa were identified in the mature brainstem. These oligomerswere sensitive to sulfhydryl-reducing agents and resistant to SDS, contrary to the situation seen in the related Na �–(K �)–Cl � cotrans-porter. In HEK-293 cells, coexpressed hemagglutinin-tagged KCC2 assembled with histidine-tagged KCC2, demonstrating formation ofhomomers. Based on these findings, we conclude that the oligomers represent KCC2 dimers, trimers, and tetramers. Finally, immunoblotanalysis identified a development-dependent increase in the oligomer/monomer ratio from embryonic day 18 to P30 throughout thebrain that correlates with KCC2 activation. Together, our data indicate that the developmental shift from depolarization to hyperpolar-ization can be determined by both increased gene expression and KCC2 oligomerization.

Key words: auditory brainstem; brain development; chloride regulation; synaptic inhibition; oligomerization; cation-chloridecotransporters

IntroductionFast inhibitory neurotransmission in the CNS is mediated byGABA and glycine via ligand-gated Cl� channels. In the vastmajority of mature neurons, a low intracellular chloride concen-tration ([Cl�]i) causes Cl� influx during channel opening, re-sulting in hyperpolarization (Eccles, 1966; Kaila, 1994). In con-trast, a high [Cl�]i in immature neurons results in Cl� efflux atthe resting membrane potential (Zhang et al., 1991; Wang et al.,1994; Obrietan and van den Pol, 1995; Ehrlich et al., 1999; Kakazuet al., 1999). This efflux depolarizes the neurons and can even

elicit action potentials. In all CNS regions analyzed so far, theneuron-specific K�–Cl� cotransporter KCC2 (Payne et al.,1996) is responsible for Cl� extrusion and, hence, for the gener-ation of the Cl�-mediated hyperpolarizing currents prevalent inmature neurons (Rivera et al., 1999; Hubner et al., 2001; Bal-akrishnan et al., 2003; Zhu et al., 2005). Recently, it was alsoshown that KCC2 is not only necessary but sufficient to causehyperpolarizing GABA currents (Chudotvorova et al., 2005; Fiu-melli et al., 2005; Lee et al., 2005). In agreement with its essentialrole for inhibitory synapses, KCC2 knock-out mice die perina-tally (Hubner et al., 2001).

KCC2 belongs to the cation– chloride cotransporter (CCC)family, which comprises the Na�–(K�)–Cl� cotransporters[N(K)CCs] NCC, NKCC1, and NKCC2, as well as the K�–Cl�

cotransporters (KCCs) KCC1, KCC2, KCC3, and KCC4 (Delpireand Mount, 2002; Payne et al., 2003). All members likely contain12 transmembrane domains (Gerelsaikhan and Turner, 2000).Biochemical studies have demonstrated a homodimeric organi-zation of NKCC1 (Moore-Hoon and Turner, 2000), NCC (deJong et al., 2003), and NKCC2 (Starremans et al., 2003). A homo-oligomeric organization was also proposed for KCCs, because atruncated KCC1 form was found to be a dominant-negative in-hibitor of KCCs (Casula et al., 2001).

Recent results point to two rate-limiting steps of KCC2 acti-

Received Jan. 16, 2006; revised Aug. 24, 2006; accepted Aug. 24, 2006.This work was supported by Deutsche Forschungsgemeinschaft Grants GRK 845/1 and 428/2-1. We thank Dr. W.

Nastainczyk for generating the rabbit anti-nKCC2 antibody, Dr. D. Bruns for the generous gift of the anti-vesicle-associated membrane protein 2 antibody, Dr. C. Hubner for providing anti-KCC2 antibody, S. Khirug for assistance atthe uncaging setup, Drs. J. W. Deitmer and C. Lohr for assistance with confocal microscopy, and Drs. R. Zimmermann,K. Kaila, and C. Rivera for helpful discussions. K. Ociepka is acknowledged for expert technical assistance and X.Futterer for help with the HEK-293 cell experiments. The monoclonal antibodies T4 from Dr. C. Lytle and �5 from Dr.Fambrough were obtained through the Developmental Studies Hybridoma Bank, developed under the auspices ofthe National Institute of Child Health and Human Development and maintained by The University of Iowa, Depart-ment of Biological Sciences, Iowa City, IA.

Correspondence should be addressed to Hans Gerd Nothwang, Abteilung Tierphysiologie, Fachbereich Biologie,Technische Universitat Kaiserslautern, Postfach 3049, D-67653 Kaiserslautern, Germany. E-mail:[email protected].

DOI:10.1523/JNEUROSCI.3257-06.2006Copyright © 2006 Society for Neuroscience 0270-6474/06/2610407-13$15.00/0

The Journal of Neuroscience, October 11, 2006 • 26(41):10407–10419 • 10407

vation during development. In the rat hippocampus, the pureincrease in the protein amount renders GABAergic action hyper-polarizing (Rivera et al., 1999). This is in contrast to the lateralsuperior olive (LSO), a conspicuous auditory structure in themedullary brainstem. Expression analysis demonstrated abun-dant KCC2 protein in immature LSO neurons without an sub-stantial increase in KCC2 during maturation (Balakrishnan et al.,2003). However, a genetic knockdown of the KCC2 geneSLC12A5 (solute carrier family 12, member 5) only affects [Cl�]i

in mature, but not in immature, LSO neurons (Balakrishnan etal., 2003). This implies the presence of transport-inactive KCC2protein during early development. A transport-inactive form ofKCC2 was also reported by other groups (Kelsch et al., 2001; Valeet al., 2003; Woodin et al., 2003; Fiumelli et al., 2005; Khirug et al.,2005). To characterize the underlying mechanism that leads totransport-inactive KCC2 protein, we here analyzed KCC2 in thedeveloping brainstem, beginning with electrophysiological andimmunohistochemical approaches. To gain insight into thestructure–function relationship of KCC2, we further analyzedthe structural organization of transport-active and transport-inactive KCC2 in the mature and immature brainstem, respec-tively. This analysis led us to propose a novel posttranslationalregulatory mechanism for KCC2.

Materials and MethodsAnimals were bred and housed in our animal facilities and treated incompliance with the current national Animal Protection Law. All proto-cols were approved by the regional animal care and use committees andadhered to the National Institutes of Health Guide for the Care and Use ofLaboratory Animals.

Slice preparation for electrophysiology. Rats between postnatal day 3(P3) and P12 were deeply anesthetized with pentobarbital (30 – 40 mg/kgbody weight, i.p.). Brains were removed in ice-cold standard physiolog-ical solution containing 124 mM NaCl, 3 mM KCl, 2 mM CaCl2, 25 mM

NaHCO3, 1.1 mM NaH2PO4, 2 mM MgSO4, and 10 mM D-glucose, equil-ibrated with 95% O2 and 5% CO2, pH 7.4, and 300-�m-thick coronalbrainstem slices were cut with a VT 1000S vibrating blade microtome(Leica, Wetzlar, Germany). Slices were allowed to recover for at least 1 hat 32°C in standard physiological solution (bubbled with 95% O2/5%CO2).

Assay of Cl� extrusion capacity in brainstem slices. The electrophysio-logical experiments were performed as described previously (Khirug etal., 2005). In brief, somatic recordings of LSO neurons were performed instandard physiological solution at room temperature (22–25°C) in thewhole-cell voltage-clamp configuration using an EPC 10 patch-clampamplifier (HEKA Elektronik, Lambrecht, Germany). Patch pipettes werefabricated from borosilicate glass (Harvard Apparatus, Edenbridge, UK),and their resistances were 5–7.5 M�. Composition of the patch-pipettesolution was as follows (in mM): 18 KCl, 111 K-gluconate, 0.5 CaCl2, 2NaOH, 10 D-glucose, 10 HEPES, 2 Mg-ATP, and 5 BAPTA, pH 7.3 ad-justed with KOH. In some experiments, the Cl � concentration was in-creased by using 40 mM KCl and 79 mM K-gluconate. The membranepotential was clamped at �70 mV, and all potential values were correctedfor a liquid junction potential of 10 mV. Only cells with a resting mem-brane potential more negative than �50 mV were included in the anal-ysis. The reversal potential of GABAergic currents (EGABA) was deter-mined from the current–voltage ( I–V) relationship obtained by stepwiseclamping the membrane potential at different holding values.

�-Carboxy-2-nitrobenzyl-caged GABA (Invitrogen, Karlsruhe, Ger-many) was diluted in the standard physiological solution (5 mM finalconcentration) and delivered to the vicinity of the patch-clamped cell viaa quartz syringe tip (100 �m internal diameter) using an UltraMi-croPump II syringe pump (World Precision Instruments, Sarasota, FL)at a flow rate of 1–5 �l/min. For local photolysis of caged GABA, the351–364 nm output of a continuous emission argon–ion laser (Enter-prise 653; Coherent, Palo Alto, CA) was delivered via a multimode opti-

cal fiber through an Olympus Optical (Tokyo, Japan) LUMPlanFl 60�water-immersion objective. The diameter of the uncaging spot was 10�m (Khirug et al., 2005). An electronic shutter (Uniblitz, Puchheim,Germany) was used to set the light pulse duration at 15–25 ms. To visu-alize the dendrites, a fluorescent indicator (Alexa Fluor 488, 50 �M; In-vitrogen) was dialyzed via the patch pipette, and dendrites were tracedusing a confocal microscope (Radiance 2100; Bio-Rad, Herts, UK).

To assess the efficacy of Cl � extrusion, 19 or 41 mM Cl � was added tothe pipette solution. These concentrations result in EGABA values of �50or �30 mV, respectively, calculated on the basis of the Nernst equa-tion. EGABA values were compared for currents evoked at the soma(EGABA,soma) and at the dendrite (EGABA,dendrite). If robust Cl � extrusionoccurs, i.e., if KCC2 activity is strong, this leads to more negative den-dritic EGABA compared with the somatic EGABA, which is clamped by theCl � load via the patch pipette.

Statistical analysis and curve fitting were performed using Origin soft-ware (Microcal Software, Northampton, MA). Averaged data are pre-sented as mean � SD. Statistical significance was assessed using a Stu-dent’s unpaired t test.

Production of a rabbit N-terminal KCC2 polyclonal antibody. A plasmidwas constructed that encodes a fusion protein of the maltose bindingprotein (New England Biolabs, Frankfurt, Germany) and the N-terminalresidues 1–93 of the rat KCC2 (Payne et al., 1996). The fusion protein wasproduced in the Escherichia coli strain BL21 (Stratagene, Heidelberg,Germany), purified with an amylose resin column, and injected subcu-taneously into rabbits to generate polyclonal sera. Polyclonal anti-N-terminal KCC2 polyclonal (nKCC2) antibodies were affinity purified bylinking the fusion protein to Affi-Gel 15 support (Bio-Rad, Munchen,Germany), following the protocol of the manufacturer. Specific antibodywas eluted with 0.1 M Na-citrate, pH 3.1, and neutralized with NaOH.After purification, anti-nKCC2 was concentrated in PBS (in mM: 130NaCl, 7 Na2HPO4, and 3 NaH2PO4, pH 7.4) using a 30 kDa cutoffvivaspin column (Vivascience, Hannover, Germany) and stored in ali-quots at �20°C.

Immunohistochemistry. Experiments were performed on SpragueDawley rats of both genders, aged P0 –P60, and on P12 wild-type andKCC2 knockdown mice (Woo et al., 2002). The genotyping of KCC2knockdown mice was performed according to the protocol described byWoo et al. (2002). Animals were deeply anesthetized with chloral hydrate(700 mg/kg body weight, i.p.) and perfused transcardially with PBS, fol-lowed by Zamboni’s solution (2% paraformaldehyde and 15% picric acidin 0.1 M phosphate buffer, pH 7.4). Brains were removed and stored infixative overnight in the refrigerator. After incubation in 30% sucrose/PBS for cryoprotection, 30-�m-thick coronal sections were cut throughthe brainstem, collected in 15% sucrose/PBS, thoroughly rinsed in PBS,and blocked for 1 h in 3% bovine serum albumin (BSA), 10% goat serum,and 0.3% Triton X-100 in Tris-buffered saline, pH 7.4. In addition to thenovel anti-nKCC2 antibody, a commercially available rabbit anti-cKCC2antibody (raised against residues 932–1043 of the rat KCC2; catalog #07-432; Upstate Biotechnology, Hamburg, Germany) was used in some ex-periments. This antibody corresponds to the antibody published previ-ously by Williams et al. (1999). No difference between the two KCC2antibodies was noticed throughout the experiments. The monoclonalmouse anti-MAP2 antibody against the microtubule-associated protein2 was obtained from Sigma (Taufkirchen, Germany) and used to identifyneuronal somata and dendrites. Antibodies were diluted 1:500 and addedto the blocking solution. Incubation with agitation was performed in arefrigerator for 24 h. After several rinses in PBS, sections were transferredto carrier solution (0.3% Triton X-100, 1% BSA, and 1% goat serum) andtreated with the secondary antibodies, goat anti-rabbit conjugated toAlexa Fluor 488, and goat anti-mouse conjugated to Alexa Fluor 546(both diluted 1:1000; Invitrogen). Sections were incubated overnightwith agitation in a refrigerator, rinsed again, mounted on slides, and airdried, and coverslips were mounted with Vectashield H-1400 anti-fademounting medium (Vector Laboratories, Burlingame, CA). Control ex-periments were performed by omitting primary antibodies and resultedin the absence of immunosignals. Photomicrographs of sagittal sectionsfrom P4 and P12 mouse brains were taken with an Axioskop (Zeiss,

10408 • J. Neurosci., October 11, 2006 • 26(41):10407–10419 Blaesse et al. • Development-Dependent Oligomerization of KCC2

Oberkochen, Germany) with a 5� objective and mounted together withAxio Vision software 4.2.

Confocal fluorescence microscopy. Images were taken with a confocallaser scanning microscope (LSM 510; Zeiss) equipped with argon (488nm) and helium–neon (543 nm) lasers and appropriate excitation andemission filters for maximum separation of Alexa Fluor 488 (488 nmexcitation; bandpass filter 505–550) and Alexa Fluor 546 (543 nm exci-tation; long-pass filter 560). This configuration prevented bleed-through. A 10� objective lens (Plan-Neofluar, 10�/numerical aperture0.3; Zeiss) and a 40� oil-immersion objective lens (Plan-Neofluar, 40�/numerical aperture 1.3 oil; Zeiss) were used. A pinhole size of 67 �mresulted in optical sections of �10 �m thickness (10� objective lens) and�1 �m (40� objective lens). Images of 2048 � 2048 pixels were obtainedand further processed with Zeiss LSM Image browser software 2.80. Fig-ures were prepared on a personal computer running Adobe Photoshop5.5 (Adobe Systems, San Jose, CA).

Electron microscopy. Preembedding immunoelectron microscopy wasperformed as follows: P4 and P12 rats were deeply anesthetized by amixture of 1.2% ketamine 10 and 0.13% Rompun in physiological saltsolution (100 �l/g body weight) and then perfused through the heartwith 4% paraformaldehyde and 0.5% glutaraldehyde in 0.1 M phosphatebuffer. Coronal vibratome sections, 500-�m-thick were prepared at thelevel of the LSO. Sections containing the LSO were cryoprotectedthrough an ascending series of sucrose solution (0.5, 1, 1.5, and 2 M) andthen subjected to two cycles of freeze–thaw in liquid nitrogen to aid thepenetration of immunoreagents into the cells. The sections were blockedfor 1 h in 10% normal goat serum and 0.2% BSA in PBS. Incubation ofprimary antibody (1:200; rabbit anti-KCC2) (Hubner et al., 2001) wasdone in carrier (PBS with 1% serum and 0.2% BSA) at 4°C overnight. Thesecondary antibody (1:100, goat anti-rabbit nanogold; Nanoprobes,Stony Brook, NY) was applied for 2 h and subsequently enhancedwith the HQ silver enhancement kit (Nanoprobes), followed by goldtoning with 0.05% gold chloride (Sigma) in 150 mM sodium acetatebuffer. After fixation with 1% osmium tetroxide, sections were dehy-drated in an ascending series of ethanol and embedded in Epon (Roth,Karlsruhe, Germany). The semithin sections were finally examinedwith a Zeiss EM 902.

Tissue RNA isolation and reverse transcription-PCR. Sprague Dawleyrats at P3 or P12 were deeply anesthetized with chloral hydrate and killedby decapitation. Their brains were rapidly removed, and the brainstemswere dissected and stored at �80°C. Total RNA was isolated from brain-stems by the guanidine thiocyanate method (Chomczynski and Sacchi,1987). The quality and quantity of the RNA samples were assessed by gelelectrophoresis and optical density measurements. Reverse transcriptionof total RNA (20 �g) was performed using standard protocols with ran-dom hexanucleotide priming and SuperScript II as enzyme (Invitrogen,Karlsruhe, Germany). The open reading frame of the KCC2 geneSLC12A5 was amplified in nine overlapping fragments, using the primerslisted in supplemental Table S1 (available at www.jneurosci.org as sup-plemental material). PCR was performed for 35 cycles in a total volumeof 50 �l. Denaturing was at 94°C for 30 s, annealing was at 60°C for 30 s,and elongation was at 72°C for 1 min. Ten microliters of each reactionwere loaded onto a high-resolution 5% polyacrylamide gel. After electro-phoresis, the gel was stained with silver nitrate and scanned using aGS-800 calibrated densitometer (Bio-Rad, Munchen, Germany). Single-strand conformational polymorphism (SSCP) analysis was performed asdescribed previously (Orita et al., 1989). In brief, PCR products weredenatured for 10 min at 95°C in SSCP loading buffer (95% formamideand 0.2% xylene cyanol), chilled on ice, and then loaded onto 5% poly-acrylamide gels with or without 10% glycerol. After electrophoresis, gelswere silver stained and scanned.

Immunoblot analysis. Proteins were resolved by linear 3– 8% Tris–acetate gel or 4 –12% Bis–Tris gel systems (NuPAGE; Invitrogen,Karlsruhe, Germany). Treatment of samples with sulfhydryl-reducingagents or detergents was for 5 min at 40°C, if not indicated otherwise. Gelloading was done in a buffer containing 0.5% lithium dodecyl sulfate(LDS) to solubilize KCC2. Electrophoretic separation lasted either 1.5(normal run) or 3 h (long run). Proteins were electrophoretically trans-ferred to polyvinylidene difluoride (PVDF) membranes (Roth) in Nu-

PAGE transfer buffer (in mM: 25 Bicine, 25 Bis-Tris, 1 EDTA, and 0.05chlorobutanol) using the NuPAGE system. PVDF-bound proteins werevisualized by staining with ponceau S. Membranes were incubated inTTBS/milk (20 mM Tris, 150 mM NaCl, 0.1% Tween 20, and 5% nonfatdry milk, pH 7.5) with the respective antiserum overnight in the refrig-erator, followed by 2 h at room temperature. Anti-nKCC2 was used forKCC2 and the monoclonal antibody T4 against NKCC1 (Lytle et al.,1995) and �5 against the �-subunit of the Na �/K �-ATPase (Lebovitz etal., 1989). T4 and �5 were obtained from the Developmental StudiesHybridoma Bank of the University of Iowa (Iowa City, IA). Anti-hemagglutinin (HA) was from Covance (Princeton, NJ) and anti-histidine (His) from Santa Cruz Biotechnology (Heidelberg, Germany).Both antibodies were used at a 1:1000 dilution. After four washes inTTBS, the secondary antibody (1:5000, goat anti-rabbit IgG or anti-mouse IgG horseradish peroxidase conjugated; Amersham Biosciences,Freiburg, Germany) was applied for 1 h in TTBS/milk. After four washesin TTBS, bound antibodies were detected using an enhanced chemilu-minescence kit and a VersaDoc 3000 documentation system (Bio-Rad,Munchen, Germany). Quantification of the volume (i.e., the product ofarea and intensity) of different KCC2-immunoreactive bands on immu-noblots was performed using the Quantity One software (Bio-Rad,Munchen, Germany). To calculate the relative amount of monomeric oroligomeric KCC2, the volume of the monomeric structure or the sum ofthe volumes of the three oligomeric structures, respectively, was dividedby the sum of the volumes of all KCC2-immunoreactive bands and mul-tiplied by 100. To get the monomer/oligomer ratio, the volume of themonomeric structure was divided by the sum of the volumes of the threeoligomeric structures. Averaged data are presented as mean � SD. Sta-tistical significance was assessed using a Student’s unpaired t test.

Deglycosylation with endoglycosidase H and N-glycosidase F. Deglyco-sylation with endoglycosidase H (endo H) and N-glycosidase F (PGNaseF) (New England Biolabs) was performed with purified membrane pro-tein fractions according to the protocol provided by the manufacturer.Membrane proteins were first incubated for 10 min at 100°C in glyco-protein denaturing buffer (0.5% SDS and 1% �-mercaptoethanol) andthen incubated for 60 min at 37°C in either 50 mM sodium citrate, pH 5.5,at 37°C supplemented with 1250 U of endo H or in 50 mM sodiumphosphate, pH 7.5, 1% NP-40, supplemented with 1250 U of PGNase.After incubation, samples were separated on a 4 –12% Bis–Tris NuPAGEsystem.

DNA constructs and His tag pull-down. The full-length cDNA encodingthe rat KCC2 was subcloned into the pENTR/D-TOPO gateway vector(Invitrogen, Karlsruhe, Germany). After sequence confirmation, theclone was transferred to the pCDNA3 vector (Invitrogen, Karlsruhe,Germany) to express wild-type KCC2 or to the pMTH2SM–HA clone(Kaufman et al., 1989) to obtain HA-tagged KCC2. To generate His10-KCC2, the open reading frame of KCC2 was amplified using a forwardprimer encoding a start methionin, His10, and amino acids 2– 6 of KCC2(CACCATGCATCATCACCATCATCACCAT-CATCACCATCTCAACAACCTGACGGACTG) and a reverse KCC2primer encoding the last six amino acids and the stop codon and clonedin the pENTR/D-TOPO gateway vector. After sequence confirmation,the clone was transferred to the pCDNA3 vector containing a gatewaycassette.

HEK-293 cells were grown to 30% confluency in 100 mm dishes andtransiently transfected using Fugene (Roche, Mannheim, Germany). Af-ter 36 h in culture, cells were harvested and lysed in 500 �l per dish of 50mM NaH2PO4, 300 mM NaCl, 10 mM imidazol, and 0.5% Triton X-100,pH 8.0, for 10 min at 4°C. For pull-down experiments, the lysates werecleared by centrifugation at 10,000 � g for 10 min, and the supernatantwas added to 10 �l of Ni 2� resin (Qiagen, Hilden, Germany). After 2 h at4°C, the mixture was pelleted and the resin were washed three times in 50mM NaH2PO4, 300 mM NaCl, 20 mM imidazol, and 0.5% Triton X-100,pH 8.0, according to the protocol of the manufacturer. Elution was in 30�l of 50 mM NaH2PO4, 300 mM NaCl, 250 mM imidazol, and 0.5% TritonX-100, and the sample was separated by electrophoresis.

Blaesse et al. • Development-Dependent Oligomerization of KCC2 J. Neurosci., October 11, 2006 • 26(41):10407–10419 • 10409

ResultsCl � extrusion capacity in immature andmature LSO neuronsThe activity of KCC2 during early postna-tal development of LSO neurons was ad-dressed previously by applying the perfo-rated patch-clamp technique to wild-typeand KCC2 knockdown mice (Balakrish-nan et al., 2003). The 95% reduction of theKCC2 level in the knockdown mice (Wooet al., 2002) affected [Cl�]i in LSO neu-rons at P12 yet not at P3. To assess theKCC2 extrusion activity more directly, weperformed whole-cell patch-clamp re-cordings from rat LSO neurons. Under aconstant Cl� load via a somatic patch pi-pette, transport-active KCC2 generates adeclining somatodendritic Cl� gradient(Jarolimek et al., 1999; Kelsch et al., 2001;Khirug et al., 2005). We determined EGABA

at the soma and at proximal dendrites, 50�m away from the soma, by focal photol-ysis of caged GABA (Khiroug et al., 2003;Khirug et al., 2005). We observed declin-ing somatodendritic Cl� gradients at P12yet not at P3 (n � 5 for each age group).Figure 1, A and B, illustrates examples ofGABA-evoked currents and the resultingI–V relationships at P3 and P12, respec-tively. At both ages, EGABA,soma was close to�50 mV, as expected from the Nernstequation for a constant load with 19 mM

[Cl�] via the patch pipette. EGABA,dendrite

was �50 mV for the P3 neuron but morenegative (�65 mV) for the P12 neuron.Statistical analysis resulted in EGABA,soma

values of �54.2 � 4.2 mV at P3 and�56.4 � 2.3 mV at P12. EGABA,dendrite was�56.2 � 5.0 mV at P3 and �69.0 � 6.2mV at P12 (Fig. 1C). Statistically signifi-cant differences between EGABA,soma andEGABA,dendrite occurred only in the P12neurons ( p � 0.01; n � 5 for each agegroup). The results show effective Cl� ex-trusion activity at P12 and ineffective ac-tivity at P3. The decrease in the amplitudeof GABA-evoked currents between P3 andP12 is consistent with a developmental de-crease in GABAergic transmission in thesuperior olivary complex (SOC) (Kotak etal., 1998; Nabekura et al., 2004).

NKCC1 is considered to be a major source of Cl� import intoimmature neurons (Sung et al., 2000; Yamada et al., 2004) and, ifalso present in LSO neurons, might mask KCC2 activity at P3 inthese cells. To exclude this possibility, we added bumetanide tothe bath at a concentration of 10 �M, known to specifically blockNKCC1 (Russell, 2000). In addition, we raised the Cl� concen-tration in the patch pipettes to 41 mM (corresponding to an ECl�

of �30 mV) to impose the neurons to the high Cl� load observedin immature LSO neurons (Balakrishnan et al., 2003). Both val-ues, EGABA,soma of �29.2 � 2.4 mV and EGABA,dendrite of �29.7 �2.9 mV, were close to the calculated value of �30 mV and notstatistically different at P3 (Fig. 1D) ( p 0.05; n � 10). Under

the same conditions, however, the difference between EGABA,soma

and EGABA,dendrite was 12.7 � 4.0 mV in P12 LSO neurons, indi-cating robust Cl� extrusion at this age (n � 3; data not shown).These results strongly imply transport-inactive KCC2 in imma-ture LSO neurons.

Polyclonal antibody production and characterizationTo be able to analyze KCC2 during development by biochemicaland immunohistochemical assays, we generated a novel antibody(anti-nKCC2) against the 93 amino acids at the N terminus of ratKCC2. We chose the N-terminal region because the availableantibodies against KCC2 were directed against the C terminus(anti-cKCC2) (Lu et al., 1999; Williams et al., 1999). We tested

Figure 1. Analysis of Cl � extrusion capacity in the developing LSO. A, Current traces obtained from a P3 LSO neuron by focalGABA uncaging at the soma and at a proximal dendrite at different holding potentials. Horizontal bars indicate the time of theuncaging UV flash. Peak amplitudes of the GABA-evoked currents were used to generate the I–V relationship (right). EGABA was�48 and �50 mV at the soma and at the dendrite, respectively. B, Current traces (left) and EGABA determination (right) for a P12LSO neuron. Horizontal bars indicate the time of the GABA uncaging UV flash. In contrast to the P3 neuron (A), EGABA,dendrite (�65mV) was clearly more negative than EGABA,soma (�56 mV). C, At P3 (n � 5), EGABA,soma (�54.6 � 4.2 mV) was not different fromEGABA,dendrite (�56.2�5.0 mV; p 0.05). Under the same conditions, EGABA,soma at P12 (�56.4�2.3 mV) differed significantlyfrom EGABA,dendrite (�69.0 � 6.2 mV; p � 0.01; n � 5). D, When the Cl � load was increased by raising the Cl � concentration inthe patch pipette to 41 mM, this resulted in an EGABA,soma of approximately �30 mV. Bumetanide was added to the bath (10 �M

final concentration) to exclude an influence of NKCC1 on EGABA. However, under these conditions, there was no difference betweenEGABA,soma (�29.2 � 2.4 mV) and �EGABA,dendrite (�29.7 � 2.9 mV; n � 10; p 0.05) at P3. E, Schematic drawing of theexperimental setup. Focal photolytic release of caged GABA was used to determine EGABA,dendrite and EGABA,soma under constantCl � load via the patch pipette. Flashes indicate sites of GABA uncaging. If EGABA,dendrite � EGABA,soma, an effective Cl � extrusionmechanism is present.


the specificity of the antibody by immunohistochemistry andimmunoblot analysis on P12 wild-type and KCC2 knockdownmice. In wild-type mice, strong labeling was obtained in the en-tire brainstem, including the SOC nuclei (Fig. 2A). At high mag-nification, strong KCC2 immunoreactivity (IR) outlined the so-mata and proximal dendrites and resulted in diffuse labeling ofthe neuropil (Fig. 2C). In contrast, in KCC2 knockdown mice,only a faint, punctuated labeling was present (Fig. 2B), making itimpossible to identify individual neurons (Fig. 2D). Identicalresults were obtained when we used the anti-cKCC2 antibody(Fig. 2E,F). The faint immunoreactivity observed with both an-tibodies in the KCC2 knockdown mice likely represents the re-sidual 5% KCC2 protein amount in these animals, because nolabeling was obtained when the anti-KCC2 antibodies were omit-ted (Fig. 2G). Immunoblot analysis confirmed the specificity ofthe anti-nKCC2 antibody. A strong signal at the correct molecu-lar weight range of 130 –140 kDa was present in the brain mem-brane fraction of wild-type animals, whereas the signal was veryfaint in the membrane fraction of KCC2 knockdown mice (Fig.2H). A weak signal at a lower molecular weight (86 kDa) wasdetected in both wild-type and knockdown mice, indicating aslight cross-reactivity of the anti-nKCC2 antibody. These resultsdemonstrate that, despite a slight cross-reactivity, the novel anti-body anti-nKCC2 is well suited for immunohistochemical KCC2analysis.

KCC2 immunoreactivity in developing LSO neuronsTo analyze age-dependent changes in the amount and distribu-tion of KCC2 at the cellular level, we focused on rat LSO neuronsand analyzed KCC2-IR at P0, P4, P12, and P60, using anti-nKCC2. Already at P0, i.e., well before Cl� extrusion is effective,KCC2-IR densely outlined the somata and proximal dendrites. A

strong and diffuse labeling pattern wasalso present in the remaining neuropil(Fig. 3A). At P4, the labeling pattern wassimilar in that the somata and proximaldendrites were again outlined (Fig. 3B).However, the diffuse neuropil labeling wasreduced, leading to the detection ofKCC2-IR at the membrane of transverselycut dendrites. The reduction in diffuseneuropil labeling proceeded with addi-tional development, resulting in clearlyoutlined somata and dendrites by P12 andP60 (Fig. 3C,D). Throughout develop-ment, the soma interior was weakly labeled(supplemental Fig. S1, available at www.j-neurosci.org as supplemental material).Indistinguishable results were obtainedwith the anti-cKCC2 antibody, i.e., therewas no difference in the intensity ofKCC2-IR at the soma surface between im-mature and mature neurons (data notshown).

The age-related reduction of KCC2-IRin the LSO neuropil may reflect pruning ofdendrites (Rietzel and Friauf, 1998) and/orthe confinement of KCC2 to dendriticsubregions in mature LSO neurons.To distinguish between these two pos-sibilities, we performed double-immunofluorescent labeling for KCC2and the dendritic marker MAP2. At all

ages, dendrites identified by MAP2 were also KCC2 immunore-active (supplemental Fig. S2A,B, available at www.jneurosci.orgas supplemental material). Thus, the reduction of KCC2-IR inthe LSO neuropil during maturation is most likely obtained viapruning of dendrites. To corroborate the localization of KCC2 atthe plasma membrane, we performed double-labeling studieswith KCC2 and markers of the biosynthetic–secretory pathway.At both P4 and P12, we observed no colocalization of KCC2 witheither protein disulfide isomerase, a marker for the endoplas-matic reticulum (Vaux et al., 1990), or with the Golgi matrixprotein 130, a marker for the Golgi apparatus (Nakamura et al.,1995) (supplemental Fig. S2C–F, available at www.jneurosci.orgas supplemental material). Only the vesicle-associated mem-brane protein 2, a marker for several types of vesicles (Jahn andSudhof, 1999; Carvalho et al., 2004), revealed a weak colocaliza-tion in the neuropil at both ages (supplemental Fig. S2G,H, avail-able at www.jneurosci.org as supplemental material).

Together, these immunohistochemical data demonstrate thatKCC2 proteins are abundant and located in the plasma mem-brane, or in close proximity to it, in LSO neurons already at a timewhen Cl� extrusion is not yet effective. This implies that KCC2transport activity in developing LSO neurons is not regulated viaan intracellular retention mechanism.

Ultrastructural analysis of KCC2 locationTo corroborate the data obtained by confocal microscopy, wenext analyzed the location of KCC2 in immature and mature LSOneurons at the ultrastructural level. Using electron microscopyand the preembedding immunogold technique, the density anddistribution of KCC2 was found to be similar to the labelingpattern observed at the light microscopic level. In semithin sec-tions, we observed KCC2-IR at somatic and dendritic plasma

Figure 2. Analysis of anti-KCC2 antibody specificity. A–F, Anti-KCC2 antibodies directed against the N terminus (anti-nKCC2;A–D) or the C terminus (anti-cKCC2; E, F ) were applied to coronal brainstem sections of P12 wild-type (A, C, E) or KCC2 knockdown(B, D, F ) mice. A, For anti-nKCC2 in wild-type mice, a moderate IR was observed in sections containing the medial nucleus of thetrapezoid body (MNTB) and the LSO, two of the main SOC nuclei. B, Only faint IR was present in KCC2 knockdown mice. C, E,High-magnification confocal images illustrated a similar labeling pattern for anti-nKCC2 and anti-cKCC2 in �/� mice. D, F,Labeling of the presumed surface of LSO neuron somata (n) and transversely cut dendrites in the neuropil (np) is indicated byarrows and arrowheads, respectively. Both antibodies revealed only faint IR in the LSO of �/� mice. G, No immunoreactivesignal was present in the LSO of wild-type mice incubated with secondary antibodies alone. Scale bar: A, B, 200 �m; C–G, 10 �m.H, Immunoblot analysis of anti-nKCC2. Membrane fractions of brains from wild-type (left) or KCC2 knockdown (right) mice wereprobed with anti-nKCC2. A strong signal in the range of the expected molecular weight was observed in �/� mice but not in�/� mice. In addition, a weak signal was observed at lower molecular weight, independent of the genotype. 7n, Facial nerve; d,dorsal; l, lateral; Mr, relative molecular mass; �/�, wild-type mice; �/�, KCC2 knockdown mice.


membranes at both P4 and P12 (Fig. 4A,B). In high-magnification electron micrographs, we localized gold particlesalong the cytoplasmic side of the dendritic plasma membranes atboth ages (Fig. 4C,D). The distribution of the gold particles washomogenous at extrasynaptic and perisynaptic sites of excitatoryand inhibitory synapses (Fig. 4E–H). Only the postsynapticthickening was almost devoid of KCC2 molecules. The only age-related change appeared to be a higher number of gold particlesinside LSO dendrites at P4 compared with P12 (Fig. 4C,D). Thelocalization of the antibody, which was raised against the N ter-minus of KCC2 (Hubner et al., 2001), at the intracellular surfaceof the plasma membrane, is in line with the model predicting anintracellular position of both protein termini (Payne et al., 1996).

Glycosylation pattern of KCC2The KCC2-IR results obtained at the light and electron micro-scopic levels, together with our electrophysiological data, sug-gested the presence of transport-inactive KCC2 molecules at theplasma membrane before Cl� extrusion becomes effective. CCCsare glycoproteins (Payne, 1997; Williams et al., 1999; de Jong etal., 2002). To determine whether transport-inactive andtransport-active KCC2 forms differ in their glycosylation pattern,we treated isolated membrane fractions from P2 and P30 ratswith the glycosidases endo H and PGNase F. Endo H cleaves onlyhigh-mannose glycosylated forms and some hybrid oligosaccha-rides from N-linked glycoproteins (Maley et al., 1989). In con-trast, PGNase F is an amidase that cleaves high-mannose, hybrid,and complex oligosaccharides from N-linked glycoproteins (Ma-ley et al., 1989). Treatment with PGNase resulted in a molecularmass shift from �130 –140 to �120 kDa for transport-inactiveand transport-active KCC2 protein (Fig. 5A). In contrast, the�130 –140 kDa protein was insensitive to endo H at both ages(Fig. 5B). These data demonstrate that transport-inactive KCC2

displays a mature glycosylation pattern. Thus, age-related differ-ences in glycosylation are unlikely to cause KCC2 activation.

Analysis for developmentally regulated KCC2 splice variantsTransport-inactive and transport-active KCC2 forms may begenerated via the expression of age-dependent splice variants. Toaddress this possibility, we amplified the open reading frame ofthe KCC2 gene SLC12A5 in nine overlapping fragments fromcDNA of P3 and P12 rat brainstem. Separation of the PCR prod-ucts on a high-resolution polyacrylamide gel revealed no age-related difference in length (supplemental Fig. S3, available atwww.jneurosci.org as supplemental material). The productsshowed a migration pattern that fits to the calculated lengths. In anext step, we analyzed whether small variations in the transcriptsoccurred that escaped detection under conditions of normalPAGE. We performed a SSCP analysis, which is capable of detect-ing single nucleotide exchanges or minimal alterations in length(Orita et al., 1989). We found no difference for the nine frag-ments amplified from P3 or P12 brainstem cDNA (data notshown).

Structural organization of KCC2We next investigated the structural organization of KCC2, withthe aim to further explore whether the functionally relevantchanges with age occur at the protein level. Previous biochemicalanalyses had demonstrated a dimeric organization of NKCC1(Moore-Hoon and Turner, 2000), NCC (de Jong et al., 2003),and NKCC2 (Starremans et al., 2003). These dimers became de-stabilized by high concentrations of the non-ionic detergent Tri-ton X-100 (�0.4%) or the anionic detergent SDS (�0.01%). Weperformed an immunoblot analysis of the membrane fractionfrom P30 rat brainstem, separated by electrophoresis in a 4 –12%Bis–Tris gel in the absence of sulfhydryl-reducing agents. It re-vealed most KCC2-IR above 200 kDa (Fig. 6), although 0.5% LDSwas present in the loading buffer. Only a minor signal was de-tected at �130 –140 kDa, which is the molecular mass of themonomeric protein (Lu et al., 1999; Williams et al., 1999; Hubneret al., 2001). In experiments involving additional treatment of themembrane fraction with 4% Triton X-100 or 0.5% SDS for 5 minat 40°C before electrophoresis, most KCC2-IR was still detectedat molecular masses higher than the monomer, with no apparentdifference to the control (Fig. 6). Under the same conditions,however, NKCC1 was present as a monomer (data not shown).Preincubation of the membrane fraction in 1% of the non-ionicdetergent N-dodecyl-�-D-maltoside or in 0.5% of the anionicdetergent N-laurylsarcosine also caused no increase in the mono-meric KCC2 fraction (Fig. 6). Together, these data indicate thatKCC2 is present in the form of higher-molecular-mass com-plexes in the mature brainstem (P30) that are resistant to deter-gents. Monomeric KCC2 might represent unassembled protein,because the membrane fraction analyzed contains all compart-ments of the biosynthetic–secretory pathway (Guillemin et al.,2005). Furthermore, the data show that the detection of mainlymonomeric KCC2 in previous reports (Lu et al., 1999; Williamset al., 1999; Hubner et al., 2001; Balakrishnan et al., 2003) is likelyattributable to the presence of sulfhydryl-reducing agents and notto SDS in the commonly used Laemmli sample buffer.

To examine the effect of sulfhydryl-reducing agents on KCC2oligomers, we incubated the membrane fraction at increasingconcentrations of �-mercaptoethanol. We improved the resolu-tion of higher-molecular-mass KCC2 molecules by performingthe electrophoretic separation in a 3– 8% Tris–acetate gel for 3 h.In the absence of reducing agents, three additional distinct

Figure 3. KCC2-IR in rat LSO neurons during maturation. A–D, Strong KCC2-IR was presentin LSO neurons (n) at P0 (A), P4 (B), P12 (C), and P60 (D) using anti-nKCC2. Both the plasmamembrane of somata (arrows) and the plasma membrane of transversely or longitudinally cutdendrites (arrowheads) were strongly KCC2 immunoreactive. The density of KCC2-IR in theneuropil (np) decreased with age. A weak labeling in the soma interior was obvious at all ages.Scale bar: (in A) AA–D, 10 �m.


KCC2-immunoreactive bands were detected, at �270, �400,and �500 kDa (Fig. 7A). This fits very well with the calculatedmass of 270, 405, and 540 kDa for dimeric, trimeric, and tet-rameric KCC2 molecules, respectively (considering 135 kDa forthe KCC2 monomer). Increasing the concentration of�-mercaptoethanol (from 0.005 to 0.1%) resulted in the disas-sembly of the three oligomeric structures (Fig. 7A). Nevertheless,even at high concentrations of reducing agents (3%�-mercaptoethanol and 1% SDS), we observed some residualoligomeric KCC2 forms (Fig. 7A). In the absence of reducingagents, all three higher-molecular-mass bands were resistant to1% SDS (Fig. 7A).

We reasoned that some of the KCC2-IR that we observed atthe high molecular mass could have been attributable to a possi-ble interaction of KCC2 with the Na�/K�-ATPase � subunit(Ikeda et al., 2004). To address this possibility, we probed theimmunoblot with an antibody against this subunit (Lebovitz etal., 1989). All immunoreactivity against the subunit was detectedat �100 kDa, which corresponds to the molecular mass of Na�/K�-ATPase � monomers (Fig. 7B). Therefore, we reject the ideathat KCC2 monomers had formed hetero-oligomers with Na�/K�-ATPase � subunits under the conditions used.

Next, we examined whether KCC2 forms homo-oligomeric

complexes in HEK-293 cells. To this end,KCC2 was first heterologously expressedin HEK-293 cells and oligomerization wasassessed. After immunoblot analysis, fourKCC2-immunoreactive bands were de-tected with similar sizes as in the maturebrainstem, i.e., at �135, �270, �400, and�500 kDa (Fig. 8A). No KCC2-IR was ob-served in untransfected HEK-293 cells.This result shows that KCC2 can formhomo-oligomers in heterologous expres-sion systems or that HEK-293 cells expressthe interacting protein partners requiredfor forming KCC2 hetero-oligomers. Toprobe directly for KCC2 homo-oligomers,we tagged KCC2 proteins with the Hispeptide (His–KCC2) or the HA peptide(HA–KCC2) and coexpressed the corre-sponding constructs in HEK-293 cells. Asa control, HA–KCC2 was coexpressedwith untagged KCC2. Purification of His–KCC2 by Ni 2� resin resulted in the copu-rification of HA–KCC2 only when His–KCC2 was coexpressed (Fig. 8B,C). Whensubstituting His–KCC2 by untaggedKCC2, HA–KCC2 could not be isolated byNi 2� beads (Fig. 8B,C). Together, thesedata demonstrate that KCC2 can indeedform homo-oligomers. Thus, the observedKCC2 oligomers in the brainstem likelyrepresent homomeric complexes, rangingfrom dimers to tetramers.

Age-dependent oligomerization ofKCC2 in the brainstemWe next set out to explore whether oli-gomerization correlates with KCC2 trans-port activation in the developing brain-stem. To do so, we obtained the membranefraction from P1/P2 rat brainstem and

separated the sample under nonreducing conditions by electro-phoresis. Because we assumed highly reactive thiol groups in theKCC2 protein, based on the residual KCC2 oligomers observed inthe presence of 3% �-mercaptoethanol, we prepared membranesin the presence of the alkylating agent iodoacetamide to preventdisulfide bonding during lysis, membrane preparation, and elec-trophoresis. When we did so, quantitative immunoblot analysisrevealed 87 � 4% of KCC2 as monomers at P1/P2, whereas thedimeric, trimeric, and tetrameric structures summed up to 13 �4% of the total KCC2 amount (Fig. 9A,B). In contrast, when weused membranes from mature brainstem (P25–P60) prepared inthe presence of iodoacetamide, 68 � 1% of KCC2 was present asoligomers (Fig. 9A,B). Thus, the monomer/oligomer ratio de-creased from 7.10 � 2.60 at P1/P2 to 0.5 � 0.03 at P25 or older(Fig. 9C). The omission of iodoacetamide from the sample atP1/P2 led to an increased amount of oligomeric KCC2, with mo-lecular masses identical to KCC2 dimer, trimer, and tetramer(Fig. 9A). In contrast, we observed no change at P25 or older forKCC2 oligomerization when iodoacetamide was omitted (Fig.9A). These data indicate that KCC2 oligomerization in the im-mature brainstem is prevented by as yet unknown mechanisms,leading to transport-inactive KCC2.

Figure 4. Ultrastructural localization of KCC2 in LSO neurons during maturation. Location of KCC2 molecules in the LSO of P4 (A,C, E, F ) and P12 (B, D, G, H ) rats, revealed with the preembedding immunogold technique. A, B, Semithin sections with labeledneurons in the LSO. Silver-intensified gold particles were found at the presumed surface of somatic and dendritic membranes(arrows). C, D, High-magnification electron micrographs illustrated labeled dendritic profiles with boutons (asterisks) makingasymmetrical synapses (arrowheads) onto the dendrite. KCC2 was localized homogenously along the perisynaptic and extrasyn-aptic membranes at both ages. The postsynaptic thickening was almost devoid of KCC2 (arrowheads). E–H, High-magnificationultrastructural analysis showed even distribution of gold particles at perisynaptic sites (arrows) of excitatory (E, G) and inhibitory(F, H ) synapses (asterisks). n, Neuronal soma; d, dendrite. Scale bars: (in A) A, B, 30 �m; (in C) C–H, 500 �m.


KCC2-IR throughout the developing brainFinally, we addressed the question of whether the abundance oftransport-inactive KCC2 protein during the depolarizing phase islimited to the LSO or a common phenomenon. To do so, westudied KCC2-IR throughout the brain at P4 and P12 (Fig. 10).This time window was chosen because [Cl�]i is still high in mostbrain areas at P4 yet has become low by P12 [neocortex (Owens etal., 1996), hippocampus (Ben Ari et al., 1989; Rivera et al., 1999),hypothalamus (Wang et al., 2001), ventral tegmental area (Ye,2000), cerebellum (Eilers et al., 2001), SOC (Ehrlich et al., 1999;Kakazu et al., 1999; Lohrke et al., 2005)]. At P4, moderate tostrong KCC2-IR was restricted to some brain regions, namely theglomeruli and mitral cell layer of the olfactory bulb, the rostro-dorsal region of the caudate–putamen, the thalamus, the hypo-thalamus, the superior colliculus, the inferior colliculus, thepons, and the medulla oblongata, including the SOC (Fig. 10A).In the cortex and the cerebellum, KCC2-IR was present in dor-

socaudal and ventrocaudal regions, respectively. In contrast, arather uniform level of KCC2-IR was observed at P12 throughoutthe brain (Fig. 10B). In the regions displaying a high KCC2-IRalready at P4, there was no additional increase in the signal,whereas in regions with weak KCC2-IR at P4, e.g., the hippocam-pus and main parts of the cortex, the signal increased until P12(Fig. 10A,B). This suggests that transport-inactive KCC2 is notlimited to the brainstem but might be also present in other brainregions during early development.

To address whether KCC2 oligomerization is associated withthis process, we assessed the oligomerization status of KCC2 invarious CNS regions during development [at embryonic day 18(E18), P1, and P30]. In all eight regions analyzed, namely theolfactory bulb, the cortex/hippocampus, the ventral telencepha-lon, the diencephalon, the tectum, the cerebellum, the brainstem,and the spinal cord, KCC2 was already detected in immunoblotsat E18 (Fig. 11). The highest amount was observed in the caudalregions (brainstem and spinal cord), whereas the lowest amountwas detected in the cortex/hippocampus. This conforms wellwith embryonic KCC2 expression reported recently (Hubner etal., 2001; Li et al., 2002; Stein et al., 2004; Wojcik et al., 2006). Ineach CNS region analyzed, monomeric KCC2 was prevailing;oligomeric KCC2 was restricted to the olfactory bulb, the cere-bellum, and the brainstem (Fig. 11). At P1, the prevailing formwas still monomeric KCC2, but now, KCC2 oligomers wereclearly present in all regions except the cortex/hippocampus,which, moreover, again displayed the lowest KCC2 abundanceamong the various regions. Finally, at P30, all regions displayedsimilar KCC2 abundance with more oligomeric than monomericKCC2. Together, these data demonstrate that development-dependent KCC2 oligomerization is not restricted to the brain-stem but that it occurs throughout the CNS.

DiscussionThere are three main conclusions of the present study: (1) func-tional KCC2 consists of disulfide-bonded homo-oligomers, (2)age-dependent KCC2 oligomerization occurs throughout thenervous system, and (3), both transport-inactive and transport-active KCC2 are present at the plasma membrane.

Figure 5. A, B, Glycosylation pattern of KCC2 during maturation. Membrane fractions fromP2 or P30 rat brainstems were left untreated or treated with PGNase F (A) or endo H (B) and thenseparated under reducing conditions by a linear 4 –12% Bis–Tris NuPAGE system for 1.5 h. Atboth ages, KCC2 was PGNase sensitive and endo H insensitive.

Figure 6. Effects of detergents on KCC2. Membranes from P30 rat brainstems were treatedfor 5 min at 40°C with no detergent (control), 4% Triton X-100, 0.5% SDS, 1% N-dodecyl-�-D-maltoside, or 0.5% N-laurylsarcosine. After adding sample buffer containing 0.5% LDS, sampleswere separated by a linear 4 –12% Bis–Tris NuPAGE system. Immunoblot analysis revealedthat, under all conditions, KCC2 was present mainly in molecular masses higher than thatexpected for the monomer. Asterisk denotes monomeric KCC2, and the bracket depicts high-molecular-mass KCC2-IR.


Oligomeric organization of KCC2Our conclusion that the structural core unit of transport-activeKCC2 is a homo-oligomer is based on evidence obtained from aseries of experiments. First, differentially tagged KCC2 fusionproteins physically interact in a heterologous expression system.Second, the molecular masses of 270, �400, and �500 kDa atwhich we detected the majority of KCC2-IR in the mature ner-vous system correspond very well with the calculated values formultimeric KCC2. Third, all three oligomeric KCC2 structureswere disulfide bonded and resistant to detergents, thus displayingsimilar biochemical properties among each other. A homo-oligomeric organization of transport-active KCC2 is in agree-ment with previous results obtained with a KCC1 dominant mu-tant form, which reduced KCC1, KCC2, or KCC3 transportactivity in Xenopus oocytes by at least two-thirds (Casula et al.,2001). These data suggest that monomeric KCCs are transportinactive and require oligomerization to become transport active.Coimmunoprecipitation studies indeed confirmed a physical as-sociation between KCC1 and the mutant form. Moreover, mod-eling of the oligomeric stoichiometry of interaction between mu-tant and wild-type KCC1, based on relative mole fractions of

injected cRNAs for both proteins, was compatible with a dimericor tetrameric state (Casula et al., 2001). The requirement of anoligomeric KCC structure for transport activity is supported byour finding that the monomer/oligomer ratio of KCC2 decreasesduring maturation of the nervous system.

Additional evidence for KCC homo-oligomerization stemsfrom a recent analysis of KCC3. In extracts of Xenopus oocytesexpressing a KCC3Q mutant, which lacks the last 338 aminoacids, oligomeric KCC3Q was detected with approximately twicethe molecular weight of monomeric truncated KCC3Q (Howardet al., 2002). Finally, a homomeric organization of CCCs was alsoreported for NKCC1 (Moore-Hoon and Turner, 2000), NCC (deJong et al., 2003), and NKCC2 (Starremans et al., 2003). To-gether, these data suggest that CCCs form homo-oligomers sim-ilar to other transporters such as the organic anion transporterOAT1 (Hong et al., 2005) and the excitatory amino acid trans-porters EAAT1 and EAAT2 (Dehnes et al., 1998).

Figure 7. Effect of sulfhydryl-reducing agents on KCC2 oligomers. A, Membranes from P30rat brainstem were treated for 5 min at 40°C with increasing concentrations of�-mercaptoethanol, separated by a linear 3– 8% Tris–acetate NuPAGE system for 3 h, andanalyzed by immunoblot. In the absence of reducing agents, four distinct KCC2-immunoreactive bands were present, with molecular masses of �130 –140, �270, �400,and �500 kDa. Asterisk denotes the monomeric KCC2, and arrows point to KCC2 oligomers. AllKCC2 oligomers were �-mercaptoethanol sensitive yet resistant to 1% SDS. B, Analysis ofoligomeric structures for the comigration with the Na �/K �-ATPase � subunit. According to A,P30 rat brainstem was analyzed for immunoreactivity against the Na �/K �-ATPase � subunit.Under all conditions, only monomeric Na �/K �-ATPase � subunits were observed, with amolecular mass of �100 kDa.

Figure 8. Physical association between differentially tagged KCC2 proteins. A, Wild-typeKCC2 was transfected in HEK-293 cells. After expression, the membrane fraction was isolatedand separated by a 3– 8% Tris–acetate NuPAGE system for 3 h and analyzed by immunoblot.Four distinct KCC2-immunoreactive bands were present, with molecular masses of �130 –140, �270, �400, and �500 kDa, similar to the KCC2 oligomers in the P30 brainstem. NoKCC2-IR was observed in untransfected HEK-293. B1, Protein lysates from HEK-293 cells ex-pressing either HA–KCC2 and His–KCC2 (lanes 1, 2) or HA–KCC2 with untagged KCC2 (lanes 3,4) were subjected to purification by Ni 2� beads and resolved by 8% SDS-PAGE. Proteins weredetected by immunoblotting with anti-HA or anti-His6 antibody. HA–KCC2 can be isolated byNi 2� beads only in the presence of coexpressed His–KCC2. Asterisk denotes the monomericKCC2, and arrowheads indicates high-molecular-mass KCC2 immunoreactivity. Control stainingwith His tag demonstrates presence of His–KCC2 only in the Ni 2� eluate of HA–KCC2/His–KCC2double-transfected cells. B2, Schematic drawing of the experimental design. If HA–KCC2 formdimers with His–KCC2, both will be purified by Ni 2� beads. C, Immunoblot of HEK-293 cellstransfected with either His–KCC2 or HA–KCC2. Anti-KCC2 detects both fusion proteins, andanti-HA detects only HA–KCC2 and not His–KCC2. This demonstrates that the antibody is notcross-reacting.


The sensitivity of KCC2 oligomers toreducing agents and their resistance to 4%Triton X-100 or 1% SDS demonstrates di-sulfide bonding as the major type of inter-action between the monomers. This con-trasts with N(K)CCs, in whichdimerization involves noncovalent inter-actions that are sensitive to detergents suchas the non-ionic detergent Triton X-100(�0.4%) or the anionic detergent SDS(�0.01%) (Moore-Hoon and Turner, 2000;de Jong et al., 2003). In addition, we noted apopulation of KCC2 that was resistant to3% �-mercaptoethanol and 1% SDS. Theresistance may be caused by enzymes suchas transglutaminases, which can generate�-glutamyl isopeptide cross-links betweenproteins (Nemes et al., 2005). Similar ob-servations were made for the cytoskeletonprotein ezrin (Berryman et al., 1995).Thus, different types of interaction are in-volved in the oligomerization of CCCs.

Oligomerization of KCC2 correlateswith onset of Cl � extrusion inthe brainstemWe previously reported transport-inactiveKCC2 in the rat auditory brainstem at P3(Balakrishnan et al., 2003). In the present study, we used a dis-tinct electrophysiological approach based on the observation thata somatodendritic EGABA gradient is a useful indicator of efficientKCC2 transport (Jarolimek et al., 1999; Kelsch et al., 2001; Khiruget al., 2005). The results of this approach, together with thosefrom immunohistochemistry, confirmed the presence oftransport-inactive KCC2 in immature LSO neurons. Focal GABAuncaging under a constant somatic Cl� load demonstrated de-tectable dendritic Cl� extrusion only in P12 LSO neurons. P3LSO neurons did not reveal a somatodendritic EGABA gradient,implying that KCC2 is not involved in Cl� homeostasis at thisearly age and thus is transport inactive. Immunohistochemistry,using confocal and electron microscopy, confirmed a high abun-dance of KCC2 at all ages. In addition, immunoblot analysis dem-onstrated no developmental increase in KCC2 protein. This con-forms well with the abundance of KCC2 protein already at E15.5,with no additional increase beyond P3 in the mouse brainstem(Stein et al., 2004). These data are in contrast to those from arecent study that suggested that an increase in KCC2 causes theshift from depolarizing to hyperpolarizing action of glycine inLSO neurons (Shibata et al., 2004).

Our biochemical analysis revealed that the monomer/oli-gomer ratio decreases from �7 to �0.5 during maturation of thebrainstem. The oligomeric KCC2-immunoreactive signals ob-served at P1/P2 (13%) are likely attributable to some brainstemregions already containing transport-active KCC2 and displayingeffective Cl� extrusion, such as the superior paraolivary nucleus(Lohrke et al., 2005) or the respiratory center (Ikeda et al., 2004).Because functional CCCs are most likely oligomers, the lack ofeffective KCC2 transport activity in immature neurons can beexplained by the prevailing monomeric form at this stage (Fig.12). This conclusion is supported by our KCC2 expression anal-ysis in other areas of the CNS. Immunohistochemistry analysisdetected KCC2 in several CNS areas during early development.More importantly, the monomer/oligomer ratio decreased in

all regions analyzed. This demonstrates that development-dependent oligomerization is a widespread phenomenon inthe CNS. Furthermore, these data are in line with oligomer-ization being a rate-limiting step of KCC2 activation duringdevelopment.

Figure 9. KCC2 oligomerization during brainstem development. A, Membranes isolated from P2 or P30 rat brainstem wereprepared in the presence (�) or absence (�) of 8% iodoacetamide. After separation by a linear 3– 8% Tris–acetate NuPAGEsystem for 3 h, immunoblot analysis was performed. In the presence of iodoacetamide, a prominent monomeric structure andlow-abundant oligomeric structures were detected at P2. At P30, oligomeric structures were dominant. In the absence of iodoac-etamide, oligomeric KCC2 structures became more prominent at P2 but not at P30. The mass shift induced by iodoacetamide isattributable to its binding to KCC2 thiol groups. Images show representative examples of four independent experiments. B,Relative abundance of the monomeric and the oligomeric structures compared with the whole amount of KCC2 at P1/P2 and in ratsaged between P30 and P60. The optical density of the four KCC2-immunoreactive bands was quantified from densitometric scansof eight lanes from three independent preparations for rats aged P1/P2 and six lanes from two independent preparations for ratsaged between P25 and P60. Both the decrease in the monomer fraction (87 � 4 to 32 � 1%) and the increase in the oligomerfraction (13 � 4 to 68 � 1%) were highly significant ( p � 0.0001). C, The monomer/oligomer ratio decreased from 7.10 � 2.60at P1/P2 to 0.5 � 0.03 in animals aged between P25 and P60.

Figure 10. KCC2-IR throughout the developing brain. A, Montage forming a sagittal sectionof a P4 rat brain. Moderate to strong KCC2-IR was observed in various brain regions, includingthe olfactory bulb (OB), the dorsorostral border of the caudate–putamen (CPu), the thalamus(Th), the hypothalamus (Hy), the superior colliculus (SC), the inferior colliculus (IC), the pons(Po), and the medulla oblongata (MO). A dorsocaudal region of the cortex (Cx) showed alsoKCC2-IR. In the cerebellum (Cb), KCC2-IR is only weak in the rostral region but moderate in theventrocaudal region. B, At P12, KCC2-IR was moderate throughout the brain, including the hip-pocampus (Hp) and the cortex, which showed only faint IR at P4. Signal intensity in the cere-bellum was stronger than at P4, but a rostrocaudal gradient was still obvious. Scale bar: A, B, 1mm. 7, Facial nucleus; MNTB, medial nucleus of the trapezoid body; PN, pontine nuclei.


Transport-inactive KCC2 is localized at theplasma membraneThe mechanisms that give rise to predominantly monomericKCC2 in the immature brainstem are unknown. Preliminary im-munohistochemical analysis suggested the retention of KCC2 inthe biosynthetic–secretory pathway in immature LSO neurons(Balakrishnan et al., 2003). However, confocal and electron mi-croscopy clearly demonstrate KCC2 labeling at the plasma mem-brane of somata and dendrites in both immature and mature LSOneurons. This is in agreement with our deglycosylation experi-ments that identified an endo H-resistant, mature glycosylation

pattern of KCC2 already at P2. The presence of KCC2 at theplasma membrane in immature auditory brainstem neurons is inaccordance with recent studies in other auditory brainstem cen-ters, which also detected KCC2-IR at the plasma membrane inimmature and mature neurons (Lohrke et al., 2005; Vale et al.,2005). Thus, localization of the protein at the plasma membraneappears to be an insufficient criterion to distinguish transport-active from transport-inactive KCC2.

KCC2 oligomerization may be prevented by development-dependent splice variants. However, even the highly sensitiveSSCP analysis did not detect any RNA variants. This is in line withour observation that monomeric KCC2 is oligomerization com-petent, because the absence of the alkylating agent iodoacetamideresulted in oligomerization of KCC2 in isolated P1/P2 membranefractions. These data suggest that oligomerization is the defaultstate that requires a specific suppression mechanism. Anothermechanism to regulate KCC2 transport activity might be itsdevelopment-dependent presence in detergent-resistant mem-branes (DRMs). However, a biochemical analysis using TritonX-100-prepared DRMs did not reveal a redistribution of KCC2among plasma membrane microdomains. At both P0 –P2 andP30, most KCC2-IR was present in the non-DRM fraction (T. M.Kranz and H. G. Nothwang, unpublished data).

Another possibility to regulate transport activity is the devel-opmentally related (de-)phosphorylation of KCC2. This mecha-nism has already been associated with changes in KCC2 transportactivity (Kelsch et al., 2001; Kahle et al., 2005; Khirug et al., 2005;de Los et al., 2006; Gagnon et al., 2006). In fact, immunoblotanalysis in the cochlear nucleus revealed an age-dependent in-crease in tyrosine phosphorylation of KCC2 (Vale et al., 2005).Finally, it cannot be excluded that auxiliary proteins are involvedin the stabilization of monomeric and/or the formation of oligo-meric KCC2 forms (Green, 1999).

Role of transport-inactive KCC2What may be the function of transport-inactive KCC2 at theplasma membrane? Obviously, this situation allows rapid activa-

Figure 11. KCC2 oligomerization during development of the nervous system. Indicatedbrain areas were dissected at E18, P1, or P30, and the membrane fraction was prepared. Afterseparation by a linear 3– 8% Tris–acetate NuPAGE for 3 h, immunoblot analysis was performed.KCC2 is detected as early as E18 in all brain areas. During development, the monomer/oligomerratio decreases in all areas analyzed. Asterisk denotes the monomeric KCC2, and the bracketindicates high-molecular-mass KCC2 immunoreactivity.

Figure 12. Model of KCC2 activation in the LSO. In immature LSO neurons, monomeric KCC2is present in the plasma membrane. Because the monomeric KCC2 is transport inactive, LSOneurons exhibit a relative high [Cl �]i. Development-dependent oligomerization activatesKCC2, thereby lowering [Cl �]i. This change in [Cl �]i results in a developmental shift fromdepolarization to hyperpolarization (D/H-shift). Hatched ovals represent glycine receptors(GlyR), and filled ovals represent individual KCC2 polypeptides.


tion of the transporter. However, the developmental shift fromdepolarizing to hyperpolarizing action of inhibitory neurotrans-mitters occurs over several days and thus does not require rapidactivation. KCC2 may, therefore, add to the growing list of bi-functional proteins, also referred to as moonlighting proteins(Jeffery, 2003; Sriram et al., 2005). Oligomerization, for instance,changes the activity of the 37 kDa subunit of glyceraldehyde-3-phosphate dehydrogenase. As a tetramer, this protein convertsglyceraldehyde-3-phosphate to 1,3-diphosphoglycerate, whereasthe monomer acts as uracil-DNA glycosylase (Meyer-Siegler etal., 1991). The cystic fibrose gene product, besides being a Cl�

channel itself, regulates the activity of other ion channels, such asthe outwardly rectifying Cl� channel ORCC (Gabriel et al., 1993;Stutts et al., 1995) and the Na� channel ENaC (Johnson et al.,1995). Therefore, a role of KCC2 aside from K�–Cl� cotransportactivity is feasible. Indeed, such a function of KCC2 has beenproposed recently (Chudotvorova et al., 2005).

In summary, our results reveal that oligomers are the domi-nant structural unit of KCC2 in the plasma membrane of themature brainstem. In contrast to N(K)CCs, oligomerization ofKCC2 involves disulfide bonds. In the nervous system, oligomer-ization is age dependent, thereby adding a novel facet to the reg-ulation of KCC2 at the protein level. The localization oftransport-inactive KCC2 at the plasma membrane in immatureneurons suggests a role in addition to transport activity andpoints to a double life of this protein.

ReferencesBalakrishnan V, Becker M, Lohrke S, Nothwang HG, Guresir E, Friauf E

(2003) Expression and function of chloride transporters during develop-ment of inhibitory neurotransmission in the auditory brainstem. J Neu-rosci 23:4134 – 4145.

Ben Ari Y, Cherubini E, Corradetti R, Gaiarsa JL (1989) Giant synaptic po-tentials in immature rat CA3 hippocampal neurones. J Physiol (Lond)416:303–325.

Berryman M, Gary R, Bretscher A (1995) Ezrin oligomers are major cy-toskeletal components of placental microvilli: a proposal for their in-volvement in cortical morphogenesis. J Cell Biol 131:1231–1242.

Carvalho E, Schellhorn SE, Zabolotny JM, Martin S, Tozzo E, Peroni OD,Houseknecht KL, Mundt A, James DE, Kahn BB (2004) GLUT4 overex-pression or deficiency in adipocytes of transgenic mice alters the compo-sition of GLUT4 vesicles and the subcellular localization of GLUT4 andinsulin-responsive aminopeptidase. J Biol Chem 279:21598 –21605.

Casula S, Shmukler BE, Wilhelm S, Stuart-Tilley AK, Su WF, Chernova MN,Brugnara C, Alper SL (2001) A dominant negative mutant of theKCC1K-Cl cotransporter: both N- and C-terminal cytoplasmic domainsare required for K-Cl cotransport activity. J Biol Chem 276:41870 – 41878.

Chomczynski P, Sacchi N (1987) Single-step method of RNA isolation byacid guanidinium thiocyanate-phenol-chloroform extraction. Anal Bio-chem 162:156 –159.

Chudotvorova I, Ivanov A, Rama S, Hubner CA, Pellegrino C, Ben Ari Y,Medina I (2005) Early expression of KCC2 in rat hippocampal culturesaugments expression of functional GABA synapses. J Physiol (Lond)566:671– 679.

Dehnes Y, Chaudhry FA, Ullensvang K, Lehre KP, Storm-Mathisen J, DanboltNC (1998) The glutamate transporter EAAT4 in rat cerebellar Purkinjecells: a glutamate-gated chloride channel concentrated near the synapse inparts of the dendritic membrane facing astroglia. J Neurosci 18:3606–3619.

de Jong JC, Van Der Vliet WA, van den Heuvel LP, Willems PH, Knoers NV,Bindels RJ (2002) Functional expression of mutations in the humanNaCl cotransporter: evidence for impaired routing mechanisms in Gitel-man’s syndrome. J Am Soc Nephrol 13:1442–1448.

de Jong JC, Willems PH, Mooren FJ, van den Heuvel LP, Knoers NV, BindelsRJ (2003) The structural unit of the thiazide-sensitive NaCl cotrans-porter is a homodimer. J Biol Chem 278:24302–24307.

de Los HP, Kahle KT, Rinehart J, Bobadilla NA, Vazquez N, San Cristobal P,Mount DB, Lifton RP, Hebert SC, Gamba G (2006) WNK3 bypasses thetonicity requirement for K-Cl cotransporter activation via a phosphatase-dependent pathway. Proc Natl Acad Sci USA 103:1976 –1981.

Delpire E, Mount DB (2002) Human and murine phenotypes associatedwith defects in cation-chloride cotransport. Annu Rev Physiol 64:803–843.

Eccles JC (1966) The ionic mechanisms of excitatory and inhibitory synap-tic action. Ann NY Acad Sci 137:473– 494.

Ehrlich I, Lohrke S, Friauf E (1999) Shift from depolarizing to hyperpolar-izing glycine action in rat auditory neurons is due to age-dependent Cl �

regulation. J Physiol (Lond) 520:121–137.Eilers J, Plant TD, Marandi N, Konnerth A (2001) GABA-mediated Ca 2�

signalling in developing rat cerebellar Purkinje neurones. J Physiol(Lond) 536:429 – 437.

Fiumelli H, Cancedda L, Poo MM (2005) Modulation of GABAergic trans-mission by activity via postsynaptic Ca 2�-dependent regulation of KCC2function. Neuron 48:773–786.

Gabriel SE, Clarke LL, Boucher RC, Stutts MJ (1993) CFTR and outwardrectifying chloride channels are distinct proteins with a regulatory rela-tionship. Nature 363:263–268.

Gagnon KB, England R, Delpire E (2006) Volume sensitivity of cation-Cl �

cotransporters is modulated by the interaction of two kinases: Ste20-related proline-alanine-rich kinase and WNK4. Am J Physiol Cell Physiol290:C134 –C142.

Gerelsaikhan T, Turner RJ (2000) Transmembrane topology of the secre-tory Na �-K �-2Cl � cotransporter NKCC1 studied by in vitro transla-tion. J Biol Chem 275:40471– 40477.

Green WN (1999) Ion channel assembly: creating structures that function.J Gen Physiol 113:163–170.

Guillemin I, Becker M, Ociepka K, Friauf E, Nothwang HG (2005) A sub-cellular prefractionation protocol for minute amounts of mammalian cellcultures and tissue. Proteomics 5:35– 45.

Hong M, Xu W, Yoshida T, Tanaka K, Wolff DJ, Zhou F, Inouye M, You G(2005) Human organic anion transporter hOAT1 forms homooli-gomers. J Biol Chem 280:32285–32290.

Howard HC, Mount DB, Rochefort D, Byun N, Dupre N, Lu JM, Fan XM,Song LY, Riviere JB, Prevost C, Horst J, Simonati A, Lemcke B, Welch R,England R, Zhan FQ, Mercado A, Siesser WB, George AL, McDonald MP,Bouchard JP, Mathieu J, Delpire E, Rouleau GA (2002) The K-Cl co-transporter KCC3 is mutant in a severe peripheral neuropathy associatedwith agenesis of the corpus callosum. Nat Genet 32:384 –392.

Hubner CA, Stein V, Hermans-Borgmeyer I, Meyer T, Ballanyi K, Jentsch TJ(2001) Disruption of KCC2 reveals an essential role of K-Cl cotransportalready in early synaptic inhibition. Neuron 30:515–524.

Ikeda K, Onimaru H, Yamada J, Inoue K, Ueno S, Onaka T, Toyoda H, ArataA, Ishikawa TO, Taketo MM, Fukuda A, Kawakami K (2004) Malfunc-tion of respiratory-related neuronal activity in Na �, K �-ATPase �2subunit-deficient mice is attributable to abnormal Cl � homeostasis inbrainstem neurons. J Neurosci 24:10693–10701.

Jahn R, Sudhof TC (1999) Membrane fusion and exocytosis. Annu Rev Bio-chem 68:863–911.

Jarolimek W, Lewen A, Misgeld U (1999) A furosemide-sensitive K �-Cl �

cotransporter counteracts intracellular Cl � accumulation and depletionin cultured rat midbrain neurons. J Neurosci 19:4695– 4704.

Jeffery CJ (2003) Moonlighting proteins: old proteins learning new tricks.Trends Genet 19:415– 417.

Johnson LG, Boyles SE, Wilson J, Boucher RC (1995) Normalization ofraised sodium absorption and raised calcium-mediated chloride secretionby adenovirus-mediated expression of cystic fibrosis transmembraneconductance regulator in primary human cystic fibrosis airway epithelialcells. J Clin Invest 95:1377–1382.

Kahle KT, Rinehart J, de Los HP, Louvi A, Meade P, Vazquez N, Hebert SC,Gamba G, Gimenez I, Lifton RP (2005) WNK3 modulates transport ofCl � in and out of cells: implications for control of cell volume and neu-ronal excitability. Proc Natl Acad Sci USA 102:16783–16788.

Kaila K (1994) Ionic basis of GABAA receptor channel function in the ner-vous system. Prog Neurobiol 42:489 –537.

Kakazu Y, Akaike N, Komiyama S, Nabekura J (1999) Regulation of intra-cellular chloride by cotransporters in developing lateral superior oliveneurons. J Neurosci 19:2843–2851.

Kaufman RJ, Davies MV, Pathak VK, Hershey JW (1989) The phosphory-lation state of eucaryotic initiation factor 2 alters translational efficiencyof specific mRNAs. Mol Cell Biol 9:946 –958.

Kelsch W, Hormuzdi S, Straube E, Lewen A, Monyer H, Misgeld U (2001)Insulin-like growth factor 1 and a cytosolic tyrosine kinase activate chlo-


ride outward transport during maturation of hippocampal neurons.J Neurosci 21:8339 – 8347.

Khiroug L, Giniatullin R, Klein RC, Fayuk D, Yakel JL (2003) Functionalmapping and Ca 2� regulation of nicotinic acetylcholine receptor chan-nels in rat hippocampal CA1 neurons. J Neurosci 23:9024 –9031.

Khirug S, Huttu K, Ludwig A, Smirnov S, Voipio J, Rivera C, Kaila K, KhirougL (2005) Distinct properties of functional KCC2 expression in imma-ture mouse hippocampal neurons in culture and in acute slices. EurJ Neurosci 21:899 –905.

Kotak VC, Korada S, Schwartz IR, Sanes DH (1998) A developmental shiftfrom GABAergic to glycinergic transmission in the central auditory sys-tem. J Neurosci 18:4646 – 4655.

Lebovitz RM, Takeyasu K, Fambrough DM (1989) Molecular characteriza-tion and expression of the (Na � � K �)-ATPase alpha-subunit in Dro-sophila melanogaster. EMBO J 8:193–202.

Lee H, Chen CX, Liu YJ, Aizenman E, Kandler K (2005) KCC2 expression inimmature rat cortical neurons is sufficient to switch the polarity of GABAresponses. Eur J Neurosci 21:2593–2599.

Li H, Tornberg J, Kaila K, Airaksinen MS, Rivera C (2002) Patterns ofcation-chloride cotransporter expression during embryonic rodent CNSdevelopment. Eur J Neurosci 16:2358 –2370.

Lohrke S, Srinivasan G, Oberhofer M, Doncheva E, Friauf E (2005) Shiftfrom depolarizing to hyperpolarizing glycine action occurs at differentperinatal ages in superior olivary complex nuclei. Eur J Neurosci22:2708 –2722.

Lu J, Karadsheh M, Delpire E (1999) Developmental regulation of theneuronal-specific isoform of K-Cl cotransporter KCC2 in postnatal ratbrains. J Neurobiol 39:558 –568.

Lytle C, Xu JC, Biemesderfer D, Forbush III B (1995) Distribution and di-versity of Na-K-Cl cotransport proteins: a study with monoclonal anti-bodies. Am J Physiol 269:C1496 –C1505.

Maley F, Trimble RB, Tarentino AL, Plummer TH Jr (1989) Characteriza-tion of glycoproteins and their associated oligosaccharides through theuse of endoglycosidases. Anal Biochem 180:195–204.

Meyer-Siegler K, Mauro DJ, Seal G, Wurzer J, Deriel JK, Sirover MA (1991)A human nuclear uracil DNA glycosylase is the 37-kDa subunit ofglyceraldehyde-3-phosphate dehydrogenase. Proc Natl Acad Sci USA88:8460 – 8464.

Moore-Hoon ML, Turner RJ (2000) The structural unit of the secretoryNa �-K �-2Cl � cotransporter (NKCC1) is a homodimer. Biochemistry39:3718 –3724.

Nabekura J, Katsurabayashi S, Kakazu Y, Shibata S, Matsubara A, Jinno S,Mizoguchi Y, Sasaki A, Ishibashi H (2004) Developmental switch fromGABA to glycine release in single central synaptic terminals. Nat Neurosci7:17–23.

Nakamura N, Rabouille C, Watson R, Nilsson T, Hui N, Slusarewicz P, KreisTE, Warren G (1995) Characterization of a cis-Golgi matrix protein,GM130. J Cell Biol 131:1715–1726.

Nemes Z, Petrovski G, Fesus L (2005) Tools for the detection and quantita-tion of protein transglutamination. Anal Biochem 342:1–10.

Obrietan K, van den Pol AN (1995) GABA neurotransmission in the hypo-thalamus: developmental reversal from Ca 2� elevating to depressing.J Neurosci 15:5065–5077.

Orita M, Suzuki Y, Sekiya T, Hayashi K (1989) Rapid and sensitive detectionof point mutations and DNA polymorphisms using the polymerase chainreaction. Genomics 5:874 – 879.

Owens DF, Boyce LH, Davis MB, Kriegstein AR (1996) Excitatory GABAresponses in embryonic and neonatal cortical slices demonstrated bygramicidin perforated-patch recordings and calcium imaging. J Neurosci16:6414 – 6423.

Payne JA (1997) Functional characterization of the neuronal-specific K-Clcotransporter: implications for [K �]o regulation. Am J Physiol273:C1516 –C1525.

Payne JA, Stevenson TJ, Donaldson LF (1996) Molecular characterizationof a putative K-Cl cotransporter in rat brain. A neuronal-specific isoform.J Biol Chem 271:16245–16252.

Payne JA, Rivera C, Voipio J, Kaila K (2003) Cation-chloride co-transporters in neuronal communication, development and trauma.Trends Neurosci 26:199 –206.

Rietzel H-J, Friauf E (1998) Neuron types in the rat lateral superior olive anddevelopmental changes in the complexity of their dendritic arbors.J Comp Neurol 390:20 – 40.

Rivera C, Voipio J, Payne JA, Ruusuvuori E, Lahtinen H, Lamsa K, Pirvola U,Saarma M, Kaila K (1999) The K�/Cl� co-transporter KCC2 rendersGABA hyperpolarizing during neuronal maturation. Nature 397:251–255.

Russell JM (2000) Sodium-potassium-chloride cotransport. Physiol Rev80:211–276.

Shibata S, Kakazu Y, Okabe A, Fukuda A, Nabekura J (2004) Experience-dependent changes in intracellular Cl � regulation in developing auditoryneurons. Neurosci Res 48:211–220.

Sriram G, Martinez JA, McCabe ER, Liao JC, Dipple KM (2005) Single-genedisorders: what role could moonlighting enzymes play? Am J Hum Genet76:911–924.

Starremans PG, Kersten FF, van den Heuvel LP, Knoers NV, Bindels RJ(2003) Dimeric architecture of the human bumetanide-sensitive Na-K-Cl co-transporter. J Am Soc Nephrol 14:3039 –3046.

Stein V, Hermans-Borgmeyer I, Jentsch TJ, Hubner CA (2004) Expressionof the KCl cotransporter KCC2 parallels neuronal maturation and theemergence of low intracellular chloride. J Comp Neurol 468:57– 64.

Stutts MJ, Canessa CM, Olsen JC, Hamrick M, Cohn JA, Rossier BC, BoucherRC (1995) CFTR as a cAMP-dependent regulator of sodium channels.Science 269:847– 850.

Sung KW, Kirby M, McDonald MP, Lovinger DM, Delpire E (2000) Abnor-mal GABAA receptor-mediated currents in dorsal root ganglion neuronsisolated from Na–K–2Cl cotransporter null mice. J Neurosci20:7531–7538.

Vale C, Schoorlemmer J, Sanes DH (2003) Deafness disrupts chloride trans-porter function and inhibitory synaptic transmission. J Neurosci23:7516 –7524.

Vale C, Caminos E, Martinez-Galan JR, Juiz JM (2005) Expression and de-velopmental regulation of the K �-Cl � cotransporter KCC2 in the co-chlear nucleus. Hear Res 206:107–115.

Vaux D, Tooze J, Fuller S (1990) Identification by anti-idiotype antibodiesof an intracellular membrane protein that recognizes a mammalian en-doplasmic reticulum retention signal. Nature 345:495–502.

Wang J, Reichling DB, Kyrozis A, Macdermott AB (1994) Developmentalloss of GABA- and glycine-induced depolarization and Ca 2� transients inembryonic rat dorsal horn neurons in culture. Eur J Neurosci6:1275–1280.

Wang YF, Gao XB, van den Pol AN (2001) Membrane properties underly-ing patterns of GABA-dependent action potentials in developing mousehypothalamic neurons. J Neurophysiol 86:1252–1265.

Williams JR, Sharp JW, Kumari VG, Wilson M, Payne JA (1999) Theneuron-specific K-Cl cotransporter, KCC2. Antibody development andinitial characterization of the protein. J Biol Chem 274:12656 –12664.

Wojcik SM, Katsurabayashi S, Guillemin I, Friauf E, Rosenmund C, Brose N,Rhee JS (2006) A shared vesicular carrier allows synaptic corelease ofGABA and glycine. Neuron 50:575–587.

Woo NS, Lu JM, England R, McClellan R, Dufour S, Mount DB, Deutch AY,Lovinger DM, Delpire E (2002) Hyperexcitability and epilepsy associ-ated with disruption of the mouse neuronal-specific K-Cl cotransportergene. Hippocampus 12:258 –268.

Woodin MA, Ganguly K, Poo MM (2003) Coincident pre- and postsynapticactivity modifies GABAergic synapses by postsynaptic changes in Cl �

transporter activity. Neuron 39:807– 820.Yamada J, Okabe A, Toyoda H, Kilb W, Luhmann HJ, Fukuda A (2004) Cl �

uptake promoting depolarizing GABA actions in immature rat neocorti-cal neurones is mediated by NKCC1. J Physiol (Lond) 557:829 – 841.

Ye JH (2000) Physiology and pharmacology of native glycine receptors indeveloping rat ventral tegmental area neurons. Brain Res 862:74 – 82.

Zhang L, Spigelman I, Carlen PL (1991) Development of GABA-mediated,chloride-dependent inhibition of CA1 pyramidal neurones of immaturerat hippocampal slices. J Physiol (Lond) 444:25– 49.

Zhu L, Lovinger D, Delpire E (2005) Cortical neurons lacking KCC2 expres-sion show impaired regulation of intracellular chloride. J Neurophysiol93:1557–1568.


development/plasticity/repair … · 2006-11-21 · zation of nkcc1 (moore-hoon and turner, 2000),...

Documents